ToolSite Gemma 4 E4B VRAM requirements
Gemma 4 E4B has 8.0B parameters. With an 8K-token context and one request it needs about 5.61 GB of GPU memory at Q4_K_M, 8.85 GB at FP8 and 17.0 GB at FP16/BF16. The published weights take 14.9 GB (BF16). On one 24 GB RTX 3090 / 4090 it runs at Q4_K_M with its full 128K-token context.
Open Gemma 4 E4B in the calculator
VRAM by quantization
Weights plus the KV cache for 8,192 tokens in FP16 and the runtime overhead (0.5 GB plus 10%). Each row opens the calculator with that setting. What the GGUF names mean.
| Precision | Weights | Total | Smallest setup |
|---|---|---|---|
| As published (BF16) | 14.9 GB | 17.0 GB | RTX 3090 / 4090 |
| FP16 / BF16 | 14.9 GB | 17.0 GB | RTX 3090 / 4090 |
| FP8 / INT8 | 7.45 GB | 8.85 GB | RTX 3060 |
| INT4 (AWQ / GPTQ) | 3.96 GB | 5.01 GB | RTX 3060 |
| GGUF Q8_0 | 7.91 GB | 9.36 GB | RTX 3060 |
| GGUF Q6_K | 6.11 GB | 7.38 GB | RTX 3060 |
| GGUF Q5_K_M | 5.28 GB | 6.46 GB | RTX 3060 |
| GGUF Q4_K_M | 4.51 GB | 5.61 GB | RTX 3060 |
| GGUF Q3_K_M | 3.64 GB | 4.66 GB | RTX 3060 |
| GGUF Q2_K | 3.12 GB | 4.09 GB | RTX 3060 |
KV cache at long context
4 of its 42 layers use full attention, 20 keep a sliding window of 512 tokens and 18 reuse the cache of earlier layers. Each extra token of context adds 16 KB of FP16 cache per request once the sliding windows are full. How the KV cache works.
| Context | KV cache, FP16 | KV cache, FP8 | Total at Q4_K_M |
|---|---|---|---|
| 4K tokens | 84 MB | 42 MB | 5.55 GB |
| 32K tokens | 532 MB | 266 MB | 6.03 GB |
| 128K tokens | 2.02 GB | 1.01 GB | 7.68 GB |
Which GPUs can run Gemma 4 E4B
With 8,192 tokens of context. Several GPUs means one tensor-parallel group of 2, 4 or 8 cards.
| GPU | Memory | Q4_K_M | FP8 |
|---|---|---|---|
| RTX 3060 | 12 GB | Fits on one | Fits on one |
| RTX 4060 Ti 16GB | 16 GB | Fits on one | Fits on one |
| RTX 3090 / 4090 | 24 GB | Fits on one | Fits on one |
| RTX 5090 | 32 GB | Fits on one | Fits on one |
| A100 40GB | 40 GB | Fits on one | Fits on one |
| Mac, 64 GB unified memory about 75% of it is usable by the GPU by default | 48 GB | Fits on one | Fits on one |
| L40S / RTX 6000 Ada | 48 GB | Fits on one | Fits on one |
| A100 / H100 80GB | 80 GB | Fits on one | Fits on one |
| Mac, 128 GB unified memory about 75% of it is usable by the GPU by default | 96 GB | Fits on one | Fits on one |
| H200 | 141 GB | Fits on one | Fits on one |
| B200 | 180 GB | Fits on one | Fits on one |
How fast Gemma 4 E4B writes
Tokens per second for one request with 8,192 tokens of context, estimated from memory bandwidth. A dash means it does not fit on one card. Try other settings in the speed calculator.
| Hardware | Bandwidth | Q4_K_M | FP8 |
|---|---|---|---|
| RTX 3060 12GB | 360 GB/s | 37–53 | 23–33 |
| RTX 4090 | 1,008 GB/s | 95–141 | 62–89 |
| RTX 5090 | 1,792 GB/s | 152–237 | 102–152 |
| M4 Max Mac (128 GB) | 546 GB/s | 55–79 | 35–49 |
| M3 Ultra Mac Studio (512 GB) | 819 GB/s | 79–116 | 51–73 |
| H100 SXM | 3,350 GB/s | 238–402 | 169–267 |
| H200 | 4,800 GB/s | 295–530 | 218–362 |
Longest context on one GPU
How many tokens of context fit on a single card with one request and an FP16 KV cache. "Full" means the model's whole context window fits.
| GPU | Memory | Q4_K_M | Q8_0 | FP8 |
|---|---|---|---|---|
| RTX 3060 | 12 GB | 128K (full) | 128K (full) | 128K (full) |
| RTX 4060 Ti 16GB | 16 GB | 128K (full) | 128K (full) | 128K (full) |
| RTX 3090 / 4090 | 24 GB | 128K (full) | 128K (full) | 128K (full) |
| RTX 5090 | 32 GB | 128K (full) | 128K (full) | 128K (full) |
| A100 40GB | 40 GB | 128K (full) | 128K (full) | 128K (full) |
| Mac, 64 GB unified memory | 48 GB | 128K (full) | 128K (full) | 128K (full) |
| L40S / RTX 6000 Ada | 48 GB | 128K (full) | 128K (full) | 128K (full) |
| A100 / H100 80GB | 80 GB | 128K (full) | 128K (full) | 128K (full) |
| Mac, 128 GB unified memory | 96 GB | 128K (full) | 128K (full) | 128K (full) |
| H200 | 141 GB | 128K (full) | 128K (full) | 128K (full) |
| B200 | 180 GB | 128K (full) | 128K (full) | 128K (full) |
Model details
- Parameters
- 8.0B (7,996,156,490)
- Layers
- 4 of its 42 layers use full attention, 20 keep a sliding window of 512 tokens and 18 reuse the cache of earlier layers
- Attention cache
- 2 KV heads × 512
- Sliding-window layers
- 2 KV heads × 256
- Context length
- 131,072 tokens
- Published weights
- 14.9 GB (BF16)
- On Hugging Face
- google/gemma-4-E4B-it
Other models
- DeepSeek V4 Flash 0731
- Qwen3.6 35B-A3B
- Qwen3.6 27B
- Qwen3.8 27B
- Qwen3.8 Flash Next
- Qwen3.8 2.4T-A95B
- Qwen3.5 9B
- Qwen3.5 122B-A10B
- Qwen3-Coder-Next
- DeepSeek V4.1 Flash
- DeepSeek V4 Flash
- DeepSeek V4 Pro
- DeepSeek V3.2
- GLM-5.3
- GLM-5.3 Flash
- GLM-5.2
- GLM-4.7 Flash
- Gemma 4 31B
- Gemma 4 26B-A4B
- Gemma 4 12B
- Kimi K3
- MiniMax M3
- MiniMax M2.7
- MiMo V2.6 Flash
- MiMo V2.6 Pro
- Mistral Medium 3.5 128B
- Nemotron 3 Nano 4B
- Nemotron 3 Nano 30B-A3B
- Nemotron 3 Super 120B-A12B
- Xing 4.0 29B-A4B
- Muse Glimmer 30B
- MiniCPM5 2B
- Llama 3.1 8B
- Llama 3.1 70B
- Qwen3 8B
- Qwen3 30B-A3B
- gpt-oss-20b
- gpt-oss-120b
- DeepSeek V3 / R1
Numbers read from the model files on Hugging Face on .